A developer has benchmarked PostgreSQL performance and cost across 23 different EC2 instance types, aiming to simplify backend architecture decisions. The analysis focuses on a mixed 90/10 read/write workload, testing various instance types with `gp3-baseline` and `gp3-max` storage configurations.

The benchmark results, available via an open-source tool, allow users to input desired requests per second (RPS) and disk size to identify the most cost-efficient instance. For instance, to achieve 33,000 RPS, the m8g.large instance with gp3-baseline storage emerged as the cheapest option at approximately $82 per month, while the c8i.large offered the best RPS per dollar on x86-64 instances.

This data provides practical insights for engineers seeking to optimize PostgreSQL deployments on AWS, moving away from potentially overcomplicated systems often inspired by large tech case studies. The methodology and benchmark code are publicly available, encouraging community contributions for broader workload coverage.